//Teamstrength (version that works with higher probability)(without extra coms)

public class Teamstrength implements ELO {
    
    //Map<Integer, Double> ratings = new HashMap();

    @Override
    public List<Double> predictResult(List<Collection<Integer>> teams) {
        return teams.stream().map(t -> (t.stream().mapToDouble(x -> (Math.pow(10,ratings.get(x)/400))).sum().getAsDouble() / teams.stream().flatMap(x -> x.stream()).mapToDouble(x -> (Math.pow(10,ratings.get(x)/400))).sum().getAsDouble() )).collect(Collectors.toList());
    }
//------------
    //another implementation of predictResult to return single doubles and allow two team collections as input (no difference is made between them) only because I'm not sure how to combine them:
    //@Override
    public double predictResult(Collection<Collection<Integer>> winnerTeams, Collection<Collection<Integer>> loserTeams, Collection<Integer> team) {
        return (team.stream().mapToDouble(x -> (Math.pow(10,ratings.get(x)/400))).sum().getAsDouble() / (winnerTeams.stream().flatMap(x -> x.stream()).mapToDouble(x -> (Math.pow(10,ratings.get(x)/400))).sum().getAsDouble()+loserTeams.stream().flatMap(x -> x.stream()).mapToDouble(x -> (Math.pow(10,ratings.get(x)/400))).sum().getAsDouble()));
//------------
    @Override
    public void evaluateResult(Collection<Collection<Integer>> winnerTeams, Collection<Collection<Integer>> loserTeams) {
        //As I'm not sure how to combine winnerTeams and loserTeams I use the above implementation of predictResult instead:
        winnerTeams.stream().forEach(t -> t.stream().forEach(x -> ratings.put(x, ratings.get(x) + 32*(1-predictResult(winnerTeams, loserTeams, t))/t.size()) );
        loserTeams.stream().forEach(t -> t.stream().forEach(x -> ratings.put(x, ratings.get(x) + 32*predictResult(winnerTeams, loserTeams, t)/t.size()) );
    }
}